The present study was designed to devise and characterize an indirect or sandwich radioimmunolabeling technique for the study of lymphocyte surface receptors of immunoglobulin nature. Mouse lymphocytes from various sources were treated by the method of Shortman et al. to remove debris and damaged cells. This was an important preliminary step, as without it, little meaning could be attached to bulk scintillation counting of labeled cell suspensions, in view of the marked tendency of dead or damaged cells to adsorb protein nonspecifically. Next, cells were reacted at 0 degrees C for 30 min with graded dilutions of unlabeled rabbit antisera against defined mouse Ig chains. After washing, the cells were reacted with a sheep anti-rabbit globulin reagent labeled with (125)I, again at graded concentrations. After further washing, lymphocyte labeling was quantitated by both bulk scintillation counting and radioautography. Conditions were defined in which nonthymus-derived cells (B cells) but not thymus-derived cells (T cells) could be labeled. Most B cells displayed kappa- and micro-chains on their surface, but some also displayed alpha- and gamma(2)-chains, though in smaller amounts. When the concentration of both the first and the second reagents were raised considerably, conditions were defined under which virtually all T cells could be labeled by polyvalent antiglobulin sera, anti-kappa sera, or, with more difficulty, by anti-micro sera. A large series of control experiments confirmed the serologic specificity of this labeling. It was shown that under equivalent conditions, B cells bind 100-400 times more antiglobulin than do T cells. The theoretical implications of the results are briefly discussed. It is argued that the sandwich approach offers certain technical advantages over direct labeling procedures for further analyses of T cell receptors and for studies of receptor metabolism.